Perioperative antimicrobial prophylaxis and prevention of hepatobiliary surgical site infections

Objective: To characterize the microbiology of hepatobiliary surgical site infections (SSIs) and to explore the relationship between specific antimicrobial prophylaxis regimens and the development of SSIs. Design: Retrospective matched case-control study comparing patient, procedure, and antimicrobial prophylaxis characteristics among patients undergoing a hepatobiliary surgical procedure with and without an SSI. Setting: A tertiary referral acute-care facility. Methods: Patients undergoing procedures defined as "BILI" (bile duct, liver, or pancreas surgery) using National Healthcare Safety Network (NHSN) definitions, excluding those undergoing concomitant liver transplantation, from January 2013 through June 2016 were included in the study population. The SSIs were identified through routine infection control surveillance using NHSN definitions. All patients who developed an SSI were considered cases. Controls were selected randomly matched 2: 1 with cases based on fiscal quarter of the procedure. Logistic regression modeling was performed to explore variables associated with SSI, including antimicrobial prophylaxis received. Results: Among 975 procedures, 80 (8.2%) resulted in an SSI. Most cases involved an organism nonsusceptible to standard prophylaxis regimens, including cefazolin (68.8%), cefazolin plus metronidazole (61.3%), and ampicillin-sulbactam (52.5%). In a multivariate model, antimicrobial coverage against Enterococcus spp (aOR, 0.58; 95% confidence interval [CI], 0.17-2.04; P = .40) and against Pseudomonas spp (aOR, 2.40; 95% CI, 0.56-10.29; P =. 24) were not protective against the development of an SSI. The presence of a documented beta-lactam allergy was significantly associated with the development of an SSI (aOR, 3.54; 95% CI, 1.36-9.19; P =. 009). Conclusions: Although SSIs at the study institution were associated with pathogens nonsusceptible to the most commonly used prophylaxis regimens, broader-spectrum coverage was not associated with a reduction in SSIs.